W3C home > Mailing lists > Public > public-appformats@w3.org > October 2007

Re: [access-control] Authenticated GET requests

From: Ian Hickson <ian@hixie.ch>
Date: Tue, 23 Oct 2007 20:49:30 +0000 (UTC)
To: Anne van Kesteren <annevk@opera.com>
Cc: "WAF WG (public)" <public-appformats@w3.org>
Message-ID: <Pine.LNX.4.62.0710232046350.16360@hixie.dreamhostps.com>

On Tue, 23 Oct 2007, Anne van Kesteren wrote:
> 
> I think the idea is to limit those cross-site requests to requests that 
> don't include authentication information. That's not feasible at the 
> moment, but the argument is that he doesn't want this specification to 
> endorse that by saying that the cross-site requests should include 
> authentication information. Rather, that an authentication request 
> without authentication information is done first.

This would be a huge pain to implement on Apache.


> > What is the attack vector that is being mitigated by not allowing it? 
> > GETs are by definition supposed to be side-effect-free.
> 
> I think "supposed" and "real world" are key words here.

But why be practical here when the earlier argument (that we'll ever be 
able to stop doing this for <img>, etc) is a theoretical one? :-)

How do you envisage ever stopping it for the other types? If we're not 
stopping it for them, there is really absolutely no point in making XHR 
more of a pain.


I'm against requiring a preflight for GET requests. It's bad enough that 
we have to do one for POST.

-- 
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'
Received on Tuesday, 23 October 2007 20:49:58 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 8 January 2008 14:10:22 GMT